Provider map

The GPU and Compute Marketplaces

The compute marketplace is not one category. Hyperscalers, neoclouds, peer-to-peer GPU exchanges, decentralized networks, and rendering-specific networks solve different buyer problems.

Buyer guideSources
"Prices set by supply and demand across 40+ data centers."
Primary source excerpt:Vast.ai, accessed 2026-07-12

Key facts

Centralized clouds: capacity, compliance, and bundled services

AWS, Google Cloud, Azure, and Oracle sell GPU infrastructure inside broader cloud platforms. The reason to use them is often not the lowest H100 price. It is the surrounding enterprise machinery: IAM, private networking, procurement, compliance, regional governance, managed storage, observability, and existing data gravity.

Hyperscaler GPU prices can be hard to compare because the accelerator is embedded inside a named instance shape with vCPUs, memory, local SSD, networking, region, operating system, reservations, and transfer terms. Buyers should treat the provider calculator as the source of truth for final quotes and record the exact region and SKU.

The hyperscaler advantage is strongest when the model depends on data already in that cloud, when compliance gates matter, or when the deployment needs managed services around the GPU. The disadvantage is quota, procurement friction, and a pricing model optimized around enterprise account structures rather than short-lived experiments.

Neoclouds: GPU-first infrastructure

Lambda and CoreWeave are examples of GPU-first providers. Their pages present GPU families directly: H100, H200, B200, A100, GH200, and related cluster offerings. This makes them easier to reason about for ML teams because the bill starts from the accelerator and cluster shape rather than a general-purpose cloud catalog.

The tradeoff is that support, regions, compliance programs, networking options, and enterprise procurement can differ from hyperscaler assumptions. For many AI teams, that is still a good exchange: lower friction for GPU access, clearer per-GPU pricing, and cluster offerings tuned to training and inference.

Peer GPU marketplaces: price discovery and host risk

Vast.ai and similar marketplaces expose the market directly. Buyers compare host reliability, GPU type, region, verification tier, storage, network score, interruptibility, and live price. This is real price discovery, which is why the same GPU can have a wide price spread.

The buyer gets optionality, but also responsibility. You need to decide how much you trust a host, whether the workload can tolerate interruption, whether data is sensitive, and whether a cheaper listing actually has the disk, network, and CPU needed to finish the job.

Decentralized compute: supply aggregation, bids, and specialized networks

Akash, io.net, and Render Network belong in the decentralized-compute conversation, but they are not interchangeable. Akash is a general decentralized compute marketplace where deployments specify resources and providers bid. io.net presents itself as a DePIN-style network aggregating GPUs from independent sources for ML workloads. Render Network is mainly a decentralized GPU rendering network with pricing denominated around OctaneBench-hours.

This page treats "Render" in the concept as Render Network, not Render.com. Render.com is a managed application hosting platform, not a decentralized GPU market. That distinction is recorded in REPORT.md because the concept used the short name.

The decentralized pitch is structural: more suppliers, visible terms, and a path for idle hardware to become market supply. The caution is equally structural: production buyers still need clear security boundaries, data handling rules, SLA expectations, region controls, and proof that the marketplace can satisfy the workload at the moment it is needed.

Site Map

The compute-market landscapeThe compute-market landscape: GPU marketplaces, decentralized compute, inference pricing, and agent-native payments for AI workloads.Free GPU and inference cost toolsClient-side GPU cost, provider price comparison, and inference throughput calculators.GPU Cost EstimatorEstimate GPU rental cost from dollars per GPU-hour, hours, token volume, throughput, GPU count, and utilization.GPU Price CompareCompare dated, first-party GPU-hour examples for H100, A100, L40S, and RTX 4090 across providers.Inference Throughput Cost CalculatorEstimate rough self-hosted LLM inference cost per request from model size, context length, batch size, output tokens, and GPU hourly price.How Compute Is PricedA buyer-focused guide to GPU-hour, token, spot, reserved, storage, egress, batching, and utilization pricing in compute markets.The GPU and Compute MarketplacesA vendor-neutral map of centralized GPU clouds, neoclouds, peer marketplaces, and decentralized compute networks.GPU Cloud Price Comparison: How to Read the TableHow to compare GPU cloud prices without mixing up per-GPU rates, node prices, marketplace risk, storage, egress, and inference throughput.Inference vs Training MarketsWhy model training, fine-tuning, batch inference, and real-time inference produce different compute markets and pricing models.Agent-Native ComputeHow x402, per-call inference, and machine-readable payment flows could let software agents buy compute autonomously.Buyer Guide: Choosing GPU or Inference ComputeA practical checklist for choosing GPU cloud, marketplace, decentralized compute, or managed inference for a workload.Compute Market GlossaryDefinitions for GPU marketplace, inference pricing, decentralized compute, x402 payments, batching, and GPU cloud terms.Sources and Pricing BibliographyAnnotated sources for ComputeMarket.io, including provider pricing pages, inference API pricing, decentralized compute docs, and x402 references.Compute Market FAQAnswers to common questions about GPU marketplaces, decentralized compute, renting GPUs cheaply, inference pricing, and agent payments.